electricsheepafrica/africa-health-facilities-south-africa

---
annotations_creators:
- no-annotation
language_creators:
- found
language:
- en
license: other
multilinguality:
- monolingual
size_categories:
- 1K<n<10K
source_datasets:
- original
task_categories:
- tabular-classification
task_ids: []
tags:
- africa
- humanitarian
- hdx
- electric-sheep-africa
- health-facilities
- hxl
- zaf
pretty_name: "South Africa Healthsites"
dataset_info:
splits:
- name: train
num_examples: 2444
- name: test
num_examples: 611
---

# South Africa Healthsites

**Publisher:** Global Healthsites Mapping Project · **Source:** [HDX](https://data.humdata.org/dataset/south-africa-healthsites) · **License:** `ODbL` · **Updated:** 2025-10-15

---

## Abstract

This dataset shows the list of operating health facilities. Attributes included: Name,Nature of Facility, Activities, Lat, Long

Each row in this dataset represents tabular records. Data was last updated on HDX on 2025-10-15. Geographic scope: **ZAF**.

*Curated into ML-ready Parquet format by [Electric Sheep Africa](https://huggingface.co/electricsheepafrica).*

---

## Dataset Characteristics

| | |
|---|---|
| **Domain** | Public health |
| **Unit of observation** | Tabular records |
| **Rows (total)** | 3,055 |
| **Columns** | 23 (8 numeric, 14 categorical, 0 datetime) |
| **Train split** | 2,444 rows |
| **Test split** | 611 rows |
| **Geographic scope** | ZAF |
| **Publisher** | Global Healthsites Mapping Project |
| **HDX last updated** | 2025-10-15 |

---

## Variables

**Geographic** — `x` (range 17.8857–32.659), `y` (range -34.7948–-22.9711), `osm_type` (node, way), `loc_amenity` (hospital, pharmacy, clinic), `meta_speciality` (general, pharmacology, general;maternity;chronic) and 3 others.

**Temporal** — `changeset_timestamp`.

**Outcome / Measurement** — `addr_housenumber` (range 1.0–25166.0).

**Identifier / Metadata** — `osm_id` (range 4353198.0–13159481275.0), `loc_name` (Clicks, Dis-Chem, Dischem), `addr_postcode` (range 1.0–9992.0), `changeset_id` (range 58907.0–173274440.0), `meta_id` and 2 others.

**Other** — `completeness` (range 6.25–78.125), `meta_healthcare` (hospital, clinic, pharmacy), `meta_operator` (government, Netcare, Life Health Care), `contact_phone` ([phone], [phone], [phone]), `addr_street` (Main Road, Voortrekker Road, Main Street) and 1 others.

---

## Quick Start

```python
from datasets import load_dataset

ds = load_dataset("electricsheepafrica/africa-health-facilities-south-africa")
train = ds["train"].to_pandas()
test = ds["test"].to_pandas()

print(train.shape)
train.head()
```

---

## Schema

| Column | Type | Null % | Range / Sample Values |
|---|---|---|---|
| `x` | float64 | 42.9% | 17.8857 – 32.659 (mean 26.592) |
| `y` | float64 | 42.9% | -34.7948 – -22.9711 (mean -28.2667) |
| `osm_id` | int64 | 0.0% | 4353198.0 – 13159481275.0 (mean 3707511376.2131) |
| `osm_type` | object | 0.0% | node, way |
| `completeness` | float64 | 0.0% | 6.25 – 78.125 (mean 23.7981) |
| `loc_amenity` | object | 3.8% | hospital, pharmacy, clinic |
| `meta_healthcare` | object | 17.7% | hospital, clinic, pharmacy |
| `loc_name` | object | 14.0% | Clicks, Dis-Chem, Dischem |
| `meta_operator` | object | 73.3% | government, Netcare, Life Health Care |
| `meta_speciality` | object | 67.2% | general, pharmacology, general;maternity;chronic |
| `meta_operator_type` | object | 70.5% | private, public, government |
| `contact_phone` | object | 78.5% | [phone], [phone], [phone] |
| `meta_emergency` | object | 77.9% | yes, no |
| `addr_housenumber` | float64 | 76.6% | 1.0 – 25166.0 (mean 481.5161) |
| `addr_street` | object | 48.8% | Main Road, Voortrekker Road, Main Street |
| `addr_postcode` | float64 | 62.0% | 1.0 – 9992.0 (mean 3045.3121) |
| `addr_city` | object | 53.0% | |
| `changeset_id` | int64 | 0.0% | 58907.0 – 173274440.0 (mean 123995642.8494) |
| `changeset_version` | int64 | 0.0% | 1.0 – 21.0 (mean 2.9061) |
| `changeset_timestamp` | datetime64[ns, UTC] | 0.0% | |
| `meta_id` | object | 0.0% | |
| `esa_source` | object | 0.0% | |
| `esa_processed` | object | 0.0% | |

---

## Numeric Summary

| Column | Min | Max | Mean | Median |
|---|---|---|---|---|
| `x` | 17.8857 | 32.659 | 26.592 | 28.1019 |
| `y` | -34.7948 | -22.9711 | -28.2667 | -26.1798 |
| `osm_id` | 4353198.0 | 13159481275.0 | 3707511376.2131 | 1311143298.0 |
| `completeness` | 6.25 | 78.125 | 23.7981 | 18.75 |
| `addr_housenumber` | 1.0 | 25166.0 | 481.5161 | 56.0 |
| `addr_postcode` | 1.0 | 9992.0 | 3045.3121 | 1746.0 |
| `changeset_id` | 58907.0 | 173274440.0 | 123995642.8494 | 127922714.0 |
| `changeset_version` | 1.0 | 21.0 | 2.9061 | 2.0 |

---

## Curation

Raw data was downloaded from HDX via the CKAN API and converted to Parquet. Column names were lowercased and standardised to snake_case. Common missing-value markers (`N/A`, `null`, `none`, `-`, `unknown`, `no data`, `#N/A`) were unified to `NaN`. 14 column(s) with >80% missing values were removed: `geo_bounds_url`, `status_operational_status`, `access_hours`, `capacity_beds`, `capacity_staff`, `meta_health_amenity_type`.... 3 column(s) were cast from string to numeric or datetime based on parse-success rate (>85% threshold). The dataset was split 80/20 into train and test partitions using a fixed random seed (42) and saved as Snappy-compressed Parquet.

---

## Limitations

- Data originates from Global Healthsites Mapping Project and has not been independently validated by ESA.
- Automated cleaning cannot correct for misreported values, definitional inconsistencies, or sampling bias in the original collection.
- The following columns have >20% missing values and should be treated with caution in modelling: `x`, `y`, `meta_operator`, `meta_speciality`, `meta_operator_type`, `contact_phone`, `meta_emergency`, `addr_housenumber`....
- Refer to the [original HDX dataset page](https://data.humdata.org/dataset/south-africa-healthsites) for the publisher's own methodology notes and caveats.

---

## Citation

```bibtex
@dataset{hdx_africa_health_facilities_south_africa,
title = {South Africa Healthsites},
author = {Global Healthsites Mapping Project},
year = {2025},
url = {https://data.humdata.org/dataset/south-africa-healthsites},
note = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)}
}
```

---

*[Electric Sheep Africa](https://huggingface.co/electricsheepafrica) — Africa's ML dataset infrastructure. Lagos, Nigeria.*